Re: Why? - Haskell-Cafe - Haskell.org

newer
Re: [Haskell-cafe] Re: Why?

Re: Why?

John D. Earle

10 Dec 2009 10 Dec '09

3:30 p.m.

My intuition says that laziness and purity are distinct whereas yours says that purity is a necessary condition. This is what needs to be reconciled. I believe that everyone is thinking that lazy evaluation and strict evaluation are similar activities whereas they are profoundly different.

Reply

Sign in to reply online Use email software

Show replies by date

Eugene Kirpichov

10 Dec 10 Dec

3:38 p.m.

New subject: Why?

2009/12/10 John D. Earle :

My intuition says that laziness and purity are distinct whereas yours says that purity is a necessary condition. This is what needs to be reconciled.

Mixing impurity and laziness makes code whose behavior is too hard to understand. So, there is no theoretical reason not to mix them, but there is a practical one.

I believe that everyone is thinking that lazy evaluation and strict evaluation are similar activities whereas they are profoundly different. _______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe

-- Eugene Kirpichov Web IR developer, market.yandex.ru

Reply

Sign in to reply online Use email software

Peter Verswyvelen

4 p.m.

New subject: Why?

Yes I remember when watching Erik Meijer's videos on Channel9 he said a similar thing: "laziness in the presence of side effects makes your head explode"... I guess the recent Microsoft Rx framework for .NET (that permits impure push-based functional reactive programming with LINQ) will soon show if the average software engineer can successfully combine side effects with laziness... Imperative/OO programming is full of guidelines to help the programmer avoid side effects: "use local state as much as possible", "never use mutable globals", "use single point of definition to avoid double out of sync data", "encapsulate encapsulate encapsulate", etc On Thu, Dec 10, 2009 at 4:38 PM, Eugene Kirpichov wrote:

2009/12/10 John D. Earle :

...
My intuition says that laziness and purity are distinct whereas yours says that purity is a necessary condition. This is what needs to be reconciled.

Mixing impurity and laziness makes code whose behavior is too hard to understand. So, there is no theoretical reason not to mix them, but there is a practical one.

...
I believe that everyone is thinking that lazy evaluation and strict evaluation are similar activities whereas they are profoundly different. _______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe

-- Eugene Kirpichov Web IR developer, market.yandex.ru _______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe

Reply

Sign in to reply online Use email software

John D. Earle

4:02 p.m.

New subject: Why?

Eugene, by purity do you mean effect free? There is a subtle difference. The lack of effects makes a language functional, but this does not imply that the language is pure. -------------------------------------------------- From: "Eugene Kirpichov" Sent: 10 Thursday December 2009 0838 To: "John D. Earle" Cc: "Haskell Cafe" Subject: Re: [Haskell-cafe] Re: Why?

2009/12/10 John D. Earle :

...
My intuition says that laziness and purity are distinct whereas yours says that purity is a necessary condition. This is what needs to be reconciled.

Mixing impurity and laziness makes code whose behavior is too hard to understand. So, there is no theoretical reason not to mix them, but there is a practical one.

...
I believe that everyone is thinking that lazy evaluation and strict evaluation are similar activities whereas they are profoundly different. _______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe

-- Eugene Kirpichov Web IR developer, market.yandex.ru

Reply

Sign in to reply online Use email software

Eugene Kirpichov

4:08 p.m.

New subject: Why?

You're right. I mean referential transparency. 2009/12/10 John D. Earle :

Eugene, by purity do you mean effect free? There is a subtle difference. The lack of effects makes a language functional, but this does not imply that the language is pure.

-------------------------------------------------- From: "Eugene Kirpichov" Sent: 10 Thursday December 2009 0838 To: "John D. Earle" Cc: "Haskell Cafe" Subject: Re: [Haskell-cafe] Re: Why?

...
2009/12/10 John D. Earle :

...
My intuition says that laziness and purity are distinct whereas yours says that purity is a necessary condition. This is what needs to be reconciled.

Mixing impurity and laziness makes code whose behavior is too hard to understand. So, there is no theoretical reason not to mix them, but there is a practical one.

...
I believe that everyone is thinking that lazy evaluation and strict evaluation are similar activities whereas they are profoundly different. _______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe

-- Eugene Kirpichov Web IR developer, market.yandex.ru

_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe

-- Eugene Kirpichov Web IR developer, market.yandex.ru

Reply

Sign in to reply online Use email software

Tom Davie

3:55 p.m.

New subject: Why?

Non-strictness is not necessary for purity, but it sure gives you some nice properties... Take for example const x y = x It would be really nice for this function to have the property "always results in x no matter what you give it as it's second argument". But for a language which is strict, all instances where computing y non-terminates also non-terminate. So yes, non-strictness is very much a property you want in a language. Bob On Thu, Dec 10, 2009 at 3:30 PM, John D. Earle wrote:

My intuition says that laziness and purity are distinct whereas yours says that purity is a necessary condition. This is what needs to be reconciled.

I believe that everyone is thinking that lazy evaluation and strict evaluation are similar activities whereas they are profoundly different. _______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe

Reply

Sign in to reply online Use email software

Sebastian Sylvan

4:56 p.m.

New subject: Why?

On Thu, Dec 10, 2009 at 3:30 PM, John D. Earle wrote:

My intuition says that laziness and purity are distinct whereas yours says that purity is a necessary condition. This is what needs to be reconciled.

I think laziness requires purity to make sense. Laziness implies that the order of evaluation is highly unpredictable and depends strongly on the implementation details of libraries and such (which you may not have access to). So it's fickle. Someone adds an if statement somewhere and all of a sudden a variable gets evaluated earlier than it used to. It would be madness to write any code which depends on this unpredictable behaviour. In other words, the expressions that get evaluated lazily must not have side effects. -- Sebastian Sylvan

Reply

Sign in to reply online Use email software

MightyByte

5:03 p.m.

New subject: Why?

On Thu, Dec 10, 2009 at 11:56 AM, Sebastian Sylvan wrote:

On Thu, Dec 10, 2009 at 3:30 PM, John D. Earle wrote:

...
My intuition says that laziness and purity are distinct whereas yours says that purity is a necessary condition. This is what needs to be reconciled.

I think laziness requires purity to make sense. Laziness implies that the order of evaluation is highly unpredictable and depends strongly on the implementation details of libraries and such (which you may not have access to). So it's fickle. Someone adds an if statement somewhere and all of a sudden a variable gets evaluated earlier than it used to. It would be madness to write any code which depends on this unpredictable behaviour. In other words, the expressions that get evaluated lazily must not have side effects.

Yes, this is discussed in section 3.2 of the paper I cited earlier (http://research.microsoft.com/en-us/um/people/simonpj/papers/history-of-hask...). That paper gives some nice insight into the history of Haskell.

Reply

Sign in to reply online Use email software

Deniz Dogan

5:07 p.m.

New subject: Why?

2009/12/10 Sebastian Sylvan :

I think laziness requires purity to make sense. Laziness implies that the order of evaluation is highly unpredictable and depends strongly on the implementation details of libraries and such (which you may not have access to). So it's fickle. Someone adds an if statement somewhere and all of a sudden a variable gets evaluated earlier than it used to. It would be madness to write any code which depends on this unpredictable behaviour. In other words, the expressions that get evaluated lazily must not have side effects. -- Sebastian Sylvan

+1 This unpredictability has bit me a few times when using LINQ (which is awesome and has lazy evaluation) with C#. -- Deniz Dogan

Reply

Sign in to reply online Use email software

Ketil Malde

6:43 p.m.

New subject: Why?

Sebastian Sylvan writes:

I think laziness requires purity to make sense. Laziness implies that the order of evaluation is highly unpredictable and depends strongly on the implementation details of libraries and such

Laziness is like single-threaded concurrency.

So it's fickle. Someone adds an if statement somewhere and all of a sudden a variable gets evaluated earlier than it used to. It would be madness to write any code which depends on this unpredictable behaviour.

I'm always amazed that shared-address space multi-threaded programs work at all. -k -- If I haven't seen further, it is by standing in the footprints of giants

Reply

Sign in to reply online Use email software

5688

Age (days ago)

5688

Last active (days ago)

Download

9 comments

8 participants

tags

participants (8)

Deniz Dogan
Eugene Kirpichov
John D. Earle
Ketil Malde
MightyByte
Peter Verswyvelen
Sebastian Sylvan
Tom Davie